M ANSARI wrote:
... These situations are getting exceedingly rare ... and engines are starting to handle even those positions better. I would guess this human superiority over engines in these rare situations will also disappear as progress in hardware and software improves. Already engines are being programmed to avoid such situations.
Yes I agree with that in general.
The point is that engines will still play badly in closed positions. unless they reach depth =100 and over. Rybka 3 won't change that even on your mighty hardware. So I think we understand each other: comps are far better in open positions, but worse in closed positions. For the inbeetween I don't know, I'm not skillful enough to tell.
M ANSARI wrote:
... These situations are getting exceedingly rare ... and engines are starting to handle even those positions better. I would guess this human superiority over engines in these rare situations will also disappear as progress in hardware and software improves. Already engines are being programmed to avoid such situations.
Yes I agree with that in general.
The point is that engines will still play badly in closed positions. unless they reach depth =100 and over. Rybka 3 won't change that even on your mighty hardware. So I think we understand each other: comps are far better in open positions, but worse in closed positions. For the inbeetween I don't know, I'm not skillful enough to tell.
I do not agree.
1)You seem to assume that we are talking about fortress ideas that you need 100 plies to see that material advantage does not help but in many closed position there is no fortress.
2)Even if fortress is the only way to draw then the program may find the best move by finding that alternatives are losing more material without
searching 100 plies.
It seems clear that rybka3 is not going to have knowledge about fortress positions but this does not mean that she is going to play bad closed positions.
In order to demonstate it you need to show examples from practical games when rybka blunders and one case is not enough because I assume closed positions are positions that happen often in games
and if rybka play bad in 1% of the closed positions and play well in 99%
of them then it is misleading to say that rybka play bad in closed positions.
In order to demonstate it you need to show examples from practical games when rybka blunders and one case is not enough because I assume closed positions are positions that happen often in games
and if rybka play bad in 1% of the closed positions and play well in 99%
of them then it is misleading to say that rybka play bad in closed positions.
Uri
And what about the opposite? In order to demonstrate Rybka plays almost perfect, you have to come up with analysis of all sorts of positions showing it indeed plays perfect!
Seriously, the easiest proof the Rybka 3 non perfectness will be when Rybka 4 will be out. Rybka 4 will beat up Rybka 3, and then people will realize Rybka 3 wasn't perfect.
In order to demonstate it you need to show examples from practical games when rybka blunders and one case is not enough because I assume closed positions are positions that happen often in games
and if rybka play bad in 1% of the closed positions and play well in 99%
of them then it is misleading to say that rybka play bad in closed positions.
Uri
And what about the opposite? In order to demonstrate Rybka plays almost perfect, you have to come up with analysis of all sorts of positions showing it indeed plays perfect!
Seriously, the easiest proof the Rybka 3 non perfectness will be when Rybka 4 will be out. Rybka 4 will beat up Rybka 3, and then people will realize Rybka 3 wasn't perfect.
The main question here is about correspondence time control and even if rybka4 beat rybka3 at tournament time control it does not mean that it can do it at correspondence time control.
In order to demonstate it you need to show examples from practical games when rybka blunders and one case is not enough because I assume closed positions are positions that happen often in games
and if rybka play bad in 1% of the closed positions and play well in 99%
of them then it is misleading to say that rybka play bad in closed positions.
Uri
And what about the opposite? In order to demonstrate Rybka plays almost perfect, you have to come up with analysis of all sorts of positions showing it indeed plays perfect!
Seriously, the easiest proof the Rybka 3 non perfectness will be when Rybka 4 will be out. Rybka 4 will beat up Rybka 3, and then people will realize Rybka 3 wasn't perfect.
The main question here is about correspondence time control and even if rybka4 beat rybka3 at tournament time control it does not mean that it can do it at correspondence time control.
Uri
I expect that Rybka 4 will be able to beat Rybka 3 at correspondence time control. We shall see when Rybka 4 will be released.
Edit: Rybka 3 must be unbeatable at correspondence chess by any future engine too. So if Rybka 4 can't beat Rybka 3, then maybe Rybka 5 will, or Shredder 22 or Naum 8.0.
